Proton-conducting perovskite oxides are attractive as electrolytes for environmentally friendly electrochemical devices, giving rise to a demand for a variety of oxides. However, complex phenomena occurring during hydration present challenges for expanding the materials library. Herein, we demonstrate the accelerated discovery of a proton-conducting oxide using data-driven structure-property maps for hydration of 8613 hypothetical perovskite oxides in descriptor spaces characterized as important by gradient boosting regressors. We constructed trustworthy hydration training data sets for 65 compounds, including literature data, by performing thermogravimetry measurements on 22 perovskites. Knowledge-based target variable engineering was necessary to capture the physicochemical fundamentals of hydration and attain high accuracy for predicting proton concentration against temperature in unknown compositions extrapolated from training data sets. The model nominates the SrSnO3 host, which was not previously recognized for proton incorporation or proton conduction, and SrSn0.8Sc0.2O3-δ demonstrated proton incorporation and conduction. The results are promising for accelerating development and applications of proton-conducting oxides.
!!!All Science Journal Classification (ASJC) codes